Skip to content

Molecular modeling of the microstructure evolution during carbon fiber processing

Summary

Desai and colleagues introduce MD-CF, a coarse-grained simulator that couples kinetic Monte Carlo (kMC) bond-formation steps with classical molecular dynamics (MD) relaxation to model carbonization and graphitization of stabilized polyacrylonitrile (PAN) precursors into carbon fibers. Inputs are the initial ladder-like carbon skeleton (heteroatom detail averaged out) and physics-based reaction rates for \( \mathrm{sp}^2 \) bond formation between neighboring ladders; the workflow targets cross-sectional microstructure and transverse elastic properties, comparing against TEM-like morphology (curved sheets, hairpin motifs), X-ray diffraction patterns, and experimental transverse moduli (order 1–5 GPa).

Methods

Scope (processing stages). The model focuses on carbonization (reported roughly 1300–2000 K) and graphitization (2700–3000 K in industrial practice, per the article’s background), after stabilization has produced ladder polymers; spinning is not simulated.

Representation. Stabilized PAN is represented as periodic, infinitely long coarse-grained chains aligned along the fiber \(Z\) axis within a cell with 3D periodic boundaries. Each ladder has saturated \(\mathrm{sp}^2\) carbons and reactive carbons intended to form new \(\mathrm{sp}^2\) links between chains.

kMC + MD cycle. Graphitization proceeds in cycles: propose new bonds between eligible nearby reactive pairs subject to distance and angular alignment criteria (distance cutoff \(R_0\), improper-angle cutoff \(\theta_0\), stochastic acceptance with probability \(\eta\); Fig. 2 in the paper), then relax the network with MD. Misalignment penalties are estimated from strain energy of misoriented ladders to justify suppressing reactions between poorly aligned pairs.

Rates and temperature. Bond-formation rates follow an Arrhenius picture consistent with activation barriers influenced by separation and alignment; the paper discusses relating barrier increases to rate suppression at carbonization temperatures (e.g., order-of-magnitude estimates around 2500 K in the barrier discussion).

Property evaluation. The authors compute XRD-comparable signals from the evolved microstructures and estimate transverse moduli, interpreting low transverse stiffness with inter-sheet sliding and longitudinal texture in the idealized cell.

1 — Hybrid kMC + classical MD (not ReaxFF RMD). Engine / integrator: MD relaxation (canonical NVT/NVE-style as defined in the original MD substeps) after each kMC bond-formation MC step (the MD-CF scheme in J. Chem. Phys. 147, 224705 (2017)). PBC 3D supercell with ladder carbon CG chains; T tied to Arrhenius rates for sp\(^2\) bond formation (~1300–3000 K regimes in background and graphitization discussion). 0 GPa isotropic hydrostatic NPT control: N/A — the cited protocol focuses on internal kinetics, not a GPa-resolved Parrinello barostat study in the wiki excerpt. Timestep, full ns-scale production span, and thermostat damping: JCP Computational; this page’s extraction_quality is partial. 2 — ReaxFF / QM training: N/A (the paper explicitly contrasts to ReaxFF-based early pyrolysis work as a different modeling line). 3 — DFT as central method: N/A.

Findings

The MD-CF workflow produces cross-sectional microstructures with curved graphitic sheets and hairpin-like features consistent with microscopy narratives in the paper. Simulated XRD agrees well with experiment. Transverse moduli fall in the 1–5 GPa range, matching high-modulus fiber experiments better than some high-strength fibers (which can be higher). Higher reaction rates in the model yield more porous microstructures and lower moduli, linking kinetics to pore formation and stiffness. Comparisons to TEM-like morphology, XRD, and GPa-scale transverse moduli are in the main bullets. Extraction is only partial in frontmatter; confirm details in papers/.../Molecular modeling of the microstructure...pdf.

Limitations

The coarse-grained reaction model averages detailed gas elimination and chemistry during carbonization (contrasted with prior ReaxFF studies of early pyrolysis steps). Chains are perfectly aligned along \(Z\), focusing on transverse structure rather than full three-dimensional texture. Absolute kinetics require approximate barriers and simplified rate rules.

Relevance to group

Methodological reference for hybrid kMC/MD microstructure evolution of carbonaceous materials; distinct from ReaxFF but relevant to reactive processing workflows.

Citations and evidence anchors

  • Methods: Sections II–III (J. Chem. Phys. 147, 224705 (2017)); DOI above.